QTL mapping and candidate gene analysis of seed vigor-related traits during artificial aging in wheat (Triticum aestivum)

High vigor seeds have greater yield potential than those with low vigor; however, long-term storage leads to a decline in this trait. The objective of this study was to identify quantitative trait loci (QTLs) for seed vigor-related traits under artificial aging conditions using a high-density genetic linkage map of wheat (Triticum aestivum) and mine the related candidate genes. A doubled haploid population, derived from a cross between Hanxuan 10 × Lumai 14, was used as the experimental material. Six controlled-environment treatments were set up, i.e. the seeds were aged for 0, 24, 36, 48, 60, and 72 h at a high temperature (48 °C) and under high humidity (relative humidity 100%). Eight traits including seed germination percentage, germination energy, germination index, seedling length, root length, seedling weight, vigor index, and simple vigor index were measured. With the prolongation of artificial aging treatment, these traits showed a continuous downward trend and significant correlations were observed between most of them. A total of 49 additive QTLs for seed vigor-related traits were mapped onto 12 chromosomes (1B, 2D, 3A, 3B, 3D, 4A, 4D, 5A, 5B, 5D, 6D, and 7A); and each one accounted for 6.01–17.18% of the phenotypic variations. Twenty-five pairs of epistatic QTLs were detected on all chromosomes, except for 5D, 6A, and 7D, and each epistasis accounted for 7.35–26.06% of the phenotypic variations. Three additive QTL hot spots were found on chromosomes 5A, 5B, and 5D, respectively. 13 QTLs, QGEe5B, QGIe5B, QSLc5B, QSLd5B, QSLf5B, QRLd5B, QRLe5B, QRLf5B, QVId5B, QVIe5B, QVIf5B, QSVId5B, and QSVIe5B, were located in the marker interval AX-94643729 ~ AX-110529646 on 5B and the physical interval 707,412,449–710,959,479 bp. Genes including TRAESCS5B01G564900, TRAESCS5B01G564200, TRAESCS5B01G562600, TraesCS5B02G562700, TRAESCS5B01G561300, TRAESCS5B01G561400, and TRAESCS5B01G562100, located in this marker interval, were found to be involved in regulating the processes of carbohydrate and lipid metabolism, transcription, and cell division during the germination of aging seeds, thus they were viewed as candidate genes for seed viability-related traits. These findings provide the basis for the seed-based cloning and functional identification of related candidate genes for seed vigor.


Results
Phenotypic changes to seed vigor-related traits in a wheat DH population under different aging conditions. In the current study, eight seed vigor indexes of a wheat DH population were measured under artificial aging conditions. As the period of aging treatment increased, the eight seed vigor-related traits in the DH population showed a gradually decreasing trend (Supplementary Table S1). At 0 h of aging treatment, the VI, SVI, SL, RL, and SW values for the female parent Hanxuan 10 were significantly higher than those of the male parent Lumai 14. At 24 and 36 h after the onset of aging treatment, the other seven traits in Hanxuan 10, except RL, were significantly greater than those of Lumai 14. After 48 h of aging treatment, the VI, SVI, SL, and SW values for Hanxuan 10 were significantly or extremely significantly greater than those for Lumai 14. After 60 h of aging treatment, the GP and GE values for Hanxuan 10 were significantly greater than those for Lumai 14. After 72 h of aging treatment, the GP, GI, SW, VI, and SVI values for Hanxuan 10 were significantly higher than those for Lumai 14. Under the six treatment conditions, eight seed vigor-related traits in the DH population exhibited the phenomenon of superaffinity separation, indicating that the genes controlling each trait were widely separated within the population. The variation coefficient for each trait in the population was more than 14% and the absolute values of skewness and kurtosis coefficients for most traits were less than 1, which conformed to normal distribution (Supplementary Table S1). The absolute values of skewness coefficient and kurtosis coefficient of some characters, such as VIf and SVIf, are greater than one, indicating that there are major QTLs. Under aging treatment, the eight seed vigor-related traits of the DH population were continuously distributed (Fig. 1), indicating that each trait was controlled by multiple genes. The heritability of each seed vigor-related trait was above 81% and the heritability of a large portion of traits reached more than 95%, indicating that these eight traits were strongly affected by genetic factors (Supplementary Table S1).

Correlation of seed vigor traits in the DH population.
Under the same treatment, the GP, GE, GI, VI and SVI values were significantly or extremely correlated in the DH population, while SW, SL, and RL were significantly or extremely correlated. Under different aging conditions, most of the seed vigor-related traits were significantly or extremely significantly correlated (Fig. 2).

Epistatic QTLs.
A total of 25 pairs of epistatic QTLs were detected, which were distributed on all of the wheat chromosomes, except 5D, 6A, and 7D. Their LOD Scores ranged from 5.03 to 9.14, their contribution to phenotypic variation were 7.35-26.06%, and epistatic interaction effects ranged from 0.02 to 61.23 (Table 2). Among them, there were 14 pairs of epistatic QTLs with a negative interaction effects, indicating that the parent epistasis effect was smaller than the recombinant epistasis effect. Eleven pairs of epistatic QTLs had positive interaction effect values, indicating that the recombinant epistasis effects were smaller than the parent epistasis effect.
Additive QTL hotspots and candidate genes. Using the ICIM methods, three additive QTL hotspots were found. The first one was within the marker interval Xgwm156 ~ Xgwm415 on chr 5A, which contains three QTLs related to SL under different aging treatments. The second one was within marker interval Xgdm3 ~ AX-109095227 on chr 5D, which contains fifteen QTLs related to GE, GI, SL, RL, VI, and SVI under different aging treatments. The last one was within marker interval AX-94643729 ~ AX-110529646 on chr 5B, it contains thirteen QTLs related to GE, GI, SL, RL, VI, and SVI under different aging treatments (Fig. 4). By the GCIM methods, three QTLs related to SL under different aging treatments were detected on the marker Xgwm415 on chr 5A, two QTLs related to SL and VI under different aging treatments were detected  Table S2). Clearly, the QTL hotspots found by the two methods are consistent. The physical distances of the first and second hotspots are too long and the number of genes is too large, so it is difficult to predict candidate genes. For the last one, the physical distance between the two markers AX-94643729 and AX-110529646 is 707412449-710959479 bp.
According to the Chinese Wheat Complete Genome Reference Sequence (IWGSC RefSeqv1.0) published by the International Wheat Whole Genome Sequencing Consortium, a total of 56 genes were found in this marker range and gene annotation was carried out by referring to the website https ://plant s.ensem bl.org/index .html. It was found that seven genes, TRAESCS5B01G564900, TRAESCS5B01G564200, TRAESCS5B01G562600, TraesCS5B02G562700, TRAESCS5B01G561300, TRAESCS5B01G561400, and TRAESCS5B01G562100 may be related to seed vigor (Table 3). GO and KEGG analyses of candidate genes. We performed GO (gene ontology) enrichment analysis on candidate genes (Supplementary Table S3, Fig. 5). The genes TraesCS5B02G561300 and TraesCS5B02G561400 appeared in four terms, and the response pathways included metabolic process, catalytic activity, cellular pro-  www.nature.com/scientificreports/ cess, and biological regulation. The gene TraesCS5B02G562100 appeared in three terms, and the response pathways involved include binding activity and membrane part. Gene TraesCS5B02G564900 appeared in two terms, the response pathways included metabolic process and biological regulation. The genes TraesCS5B02G562600, TraesCS5B02G562700 appeared in one term, and the participating response pathway was catalytic activity. KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis was performed to mine candidate genes (Supplementary Table S4, Fig. 6). Genes TraesCS5B02G562600 and TraesCS5B02G562700 were significantly enriched in five metabolic pathways of 3-phytase, alkaline phosphatase D, solute carrier family 3, cetylajmaline esterase, and cholinesterase. The gene TraesCS5B02G562100 is significantly enriched in eight metabolic pathways, which include nucleotides, protein metabolism and ribosome assembly, etc. The gene TraesCS5B02G564200 was significantly enriched in the three metabolic pathways of C-Jun-amino-terminal kinase-interacting p, serine and arginine repetitive matrix 1, and ATP-dependent (S)-NAD(P)H-hydrate dehydra. The genes TraesCS5B02G561300 Table 2. Epistatic QTLs for seed vigor traits in wheat DH population. Positive value for epistatic effect means the greater genetic values of parental types than those of recombination types, while the negative value indicates the opposite scenarios. PVE = percentage of phenotypic variance explained by each QTL, AA = Additive-by-additive interaction effect.

Discussion
Significant phenotypic correlation among wheat seed vigor-related traits. Seed vigor is affected by various external conditions. Under high temperature and humidity, a large number of seeds at different degrees of aging can be obtained, which is conducive to further research on vigor theory and applications. This method has provided useful results for investigations into the seed vigor of various crops. Ye 27 studied the response to high temperatures of rice seeds containing different water contents. Under long-term high temperature conditions, seed vigor and the longevity of high water content seeds were low. Han et al. 2 analyzed the GP and Ge of maize seeds under artificial aging and carried out relevant QTL mapping.    www.nature.com/scientificreports/ In this study, the GP, GE, GI, SL, RL, SW, VI, and SVI values of the DH population decreased continuously with the increasing period of aging treatment. There were significant differences in seed vigor-related traits between parents Hanxuan 3 and Lumai 14 of DH population under artificial aging treatment, which may be due to the great differences in the source and physiological characteristics of parents. The absolute values of skewness and kurtosis coefficient of each trait in DH population are mostly less than one, which conforms to normal distribution. The absolute values of skewness coefficient and kurtosis coefficient of some traits, such as VIf and SVIf, are greater than one, indicating that there are major QTLs. Under different aging conditions, GP, GE, GI, VI, and SVI were significantly positively correlated, while SW, SL, and RL were also significantly correlated under most aging conditions. Under different aging conditions, GP, GE, GI, VI, and SVI were significantly positively correlated, while SW, SL, and RL were also significantly correlated under most aging conditions. QTL mapping was carried out for the eight seed vigor-related traits and four additional QTL hot spots were found. First, three QTLs related to SL under different aging treatments was detected in the marker range xgwm156 ~ xgwm415 on chromosome 5A. Second, QTLs related to GE, GI, SL, RL, VI, and SVI were detected in the marker interval Xgdm3 ~ AX-109095227 on chromosome 5D under different aging treatments. Third, a QTL associated with VI and SVI was detected on chromosome 5D in marker interval AX-110091432 ~ Xgdm68. Fourth, a QTL associated with six traits including GE, GI, SL, RL, VI, and SVI was detected on chromosome 5B under different aging treatments in the marker interval AX-94643729 ~ AX-110529646. Further mapping showed that these QTLs may be due to multiple effects or close linkage, which provided good molecular genetic evidence for the significant phenotypic correlation among the traits.
New QTLs for wheat seed vigor-related traits. In recent years, great progress has been made in the QTL mapping of various important traits including seed vigor in many crops. So far, research has focused on a limited number of plants such as rice, Arabidopsis, cabbage, barley, and sorghum. For example, Shi et al. 28 used two maize recombinant inbred line populations under low temperatures to detect 26 QTLs related to seed vigor such as GP and GI. These QTLs were located on all chromosomes, except chromosome 10. However, there are few studies on QTL mapping of wheat seed vigor. Misheva et al. 29 performed QTL mapping on GP and SL of aged wheat seeds under osmotic stress. The results showed that 20 QTLs were detected and distributed on chromosomes 1D, 2D, 4D, 5D, and 7D. Arif et al. 23 detected loci associated with seed longevity in bread wheat on chromosomes 1D, 2A, 7B, and 7D. Zuo et al. 30 found that chromosomes 2D, 3D, 4A, and 6B of wheat are important related to six seed germination parameters. In this study, 49 additive QTLs of wheat seed vigor-related traits were detected and found to be distributed on 12 chromosomes including 1B, 2D, 3A, 3B, 3D, 4A, 4D, 5A, 5B, 5D, 6D, and 7A. Among them, the seed vigor-related QTLs on 8 chromosomes (1B, 3A, 3B, 4D 5A, 5B, 6D, and 7A) (Supplementary Table S5) have not previously been reported, indicating that these may be new related QTLs.
Prediction of candidate genes related to seed vigor in wheat. Seed germination is a complex physiological process involving the metabolism of carbohydrates, lipids, proteins, and other substances. In this study, 56 genes were found in the marker interval AX-94643729 ~ AX-110529646 (physical range 707,412,449-710,959,479 bp) on chromosome 5B, and 7 candidate genes related to seed vigor were screened out, according to gene function annotation and the results of GO and KEGG analyses. The functional annotation of the candidate gene TraesCS5B01G564900 is an invertase/pectin methylesterase (PME) inhibitor family protein. Hothorn et al. 31 demonstrated that PME and invertase are key enzymes for plant carbohydrate metabolism, which are involved in the development of plant roots, stems, and fruits. Inhibitors of these two enzymes form the sequence family of extracellular proteins and participate in the regulation of carbohydrate metabolism. The functional annotation of the candidate gene TraesCS5B01G562600 and TraesCS5B01G562700 are GDSL esterase/lipase protein. The GDSL lipase gene has been shown to be involved in plant growth and development, lipid metabolism, and stress responses 32 . BnLIP1, the GDSL lipase gene from Brassica napus, changes with seed germination and may also be involved in regulating other physiological processes 33 . During the germination of aging seeds, the participation of transcription factors can ensure that the target gene is expressed at a specific time in a specific space. The functional annotation of the candidate gene Traescs5b01g564200 is a Trihelix transcription factor. Studies have shown that the Trihelix transcription factor is involved in plant embryogenesis and seed development 24,34 , along with the plant abiotic stress response 35 .
Cytokinins are a class of plant hormones that promote cell division, bud formation, and growth and are related to seed development 36,37 . Jameson et al. 37 detected cytokinins in bloated pea seeds and found that they were actively synthesized during germination. In this study, the candidate genes TRAESCS5B01G561300.1 and TRAESCS5B01G561400.1 were found to be functionally related to cytokinin nucleoside 5-monophosphate phosphatase. Other factors involved in plant growth and development are members of the WD40 protein family, which act as scaffolds in biomacromolecule interactions and exist widely in eukaryotes. In Arabidopsis, these proteins are considered to be the key regulatory factors in signal transmission during development and stress. For example, the WD40 repeat sequence protein AGB1 in Arabidopsis negatively regulates the signal transduction of auxin and affects cell division in fruits, hypocotyls, and roots 38,39 . GIGANTUS1 (GTS1), a member of the WD40 protein superfamily, is highly expressed in Arabidopsis embryos and controls seed germination, growth, and biomass accumulation through interactions with ribosomal protein chaperones 40 . Nucleosome remodeling factor complex components 101 and 102, which contain WD40 repeat sequences in maize, regulate seed germination, plant height, flowering time, etc. by regulating chromatin modification 41 . The functional annotation of the candidate gene Traescs5b01g562100.1 was that of a WD40 repeat domain and the gene was also found to be closely related to seed vigor. In future studies, the expression of these seven genes during the germination of aging seeds should be analyzed to verify their functions using transgene experiments, following by investigations into the mechanisms underlying their functions. The field design consisted of randomized complete blocks with three replications. Each plot consisted of two rows of 2 m long, with 25 cm between rows. Forty seeds were sown in each row. Water and fertilizer management during the growth period was the same as that used in the local production practice. The seeds were harvested in June 2018.

Experiment methods.
The experiment was carried out at Shanxi Agricultural University between January and July, 2019. First, the seeds were artificially aged. There were six controlled-environment treatments for each line, 100 seeds for each treatment, and three replicates. Seeds were placed in an artificial climate incubator at a temperature of 48 °C and relative humidity of 100%. The seeds were aged under dark conditions for 0 h (a), 24 h (b), 36 h (c), 48 h (d), 60 h (e), and 72 h (f), respectively. The aged seeds were dried at room temperature for 3 days and then the standard germination test was carried out. The seeds were first disinfected, soaked in a 0.1% HgCl 2 solution for 15 min, and washed three times with distilled water. The seeds were placed with the ventral groove facing down evenly in a culture dish (12 cm in diameter) with two layers of filter paper. An appropriate amount of distilled water was added and the dish was put in an artificial climate incubator (25 °C, 75% relative humidity) for germination. During germination, the photoperiod was 16 h light / 8 h dark. The plates were checked regularly and distilled water was replenished with an appropriate amount every day. From the third day of culture, the germination number for each line was counted. The shortest protruding part of seed radicle was the same as the length of seed, which was regarded as normal germination. The seedling length (SL), root length (RL), and seedling weight (SW) of each line were measured on the seventh day of culture. Calculation of wheat seed vigor related traits: In the formula above, n 3 is the number of germinated seeds on the 3rd day of culture, n 7 is the number of germinated seeds on the 7th day of culture, Gt is the daily germination number, Dt is the corresponding germination days of Gt. Data analysis. Relevant t-test and analysis of variance (ANOVA) were carried out by the statistical software package SPSS 16.0, frequency distribution map was carried out by Excel 2007, phenotypic correlation analysis map and phylogenetic clustering map were created in the R-package corrplot (version 0.84) 42 and clusterProfiler (version 3.16) 43 , respectively. The heritability of the traits was analyzed by analysis of variance. The formula of heritability is: H B 2 = σ g 2 /(σ g 2 + σ e 2 /r) × 100%, in which σ g 2 is the genetic variance, σ e 2 is the random error variance, and r is the number of trials repeated. QTL mapping. The genetic map of the DH population was constructed by Jing Ruilian's team at the Institute of Crop Science, Chinese Academy of Agricultural Sciences. The map was created by integrating the same genetic position with SNP markers and SSR markers. The final number of markers was 1854, of which there were 1630 660 K SNP markers. There were 224 SSR markers, the total length was 4082.44 cM, and the average distance between the markers was 2.20 cM, including 30 linkage groups.
In recent years, the statistical methods commonly used in QTL mapping are mainly MCIM, ICIM and GCIM 44 . In this study, QTL analysis was analyzed by IciMapping software for seed vigor-related traits after artificial aging treatment, with inclusive composite interval mapping (ICIM) 45 . The average value of the three repeated measurements of seed vigor-related traits in the DH population was used as the phenotypic value. The walking speed for all QTL detections was chosen at 1.0 centimorgans (cM), with P = 0.001 in stepwise regression. Based on 2000 permutations at a probability of 0.01, the significant QTL threshold was obtained. Combined with this genetic map, a main-effect and epistatic QTL mapping were performed. At the same time, QTL IciMapping software was used to map the distribution of QTL loci on wheat chromosomes. The QTLs are named after "Q + trait name + processing time + chromosome". The GCIM method in the software QTL.gCIMapping from the R website (https ://cran.r-proje ct.org/web/packa ges/QTL.gCIMa pping /index .html) was also used to identify QTLs for the above traits, with the purpose of identifying the results from the ICIM method; the critical LOD scores for a significant QTL was also set at 2.5, and the walking speed for the genome-wide scan was set at 1 cM.
The GO and KEGG analyses. The online prediction website (https ://geneo ntolo gy.org/) was used to perform GO (Gene Ontology) enrichment analysis on candidate genes. KEGG (Kyoto Encyclopedia of Genes and Genomes) analysis was performed on candidate genes using the KEGG analysis website (https ://www.kegg.jp/ keg/kegg1 .html)) 46,47 .

Conclusion
A total of 49 additive QTLs of wheat seed vigor-related traits were detected and distributed on 12 chromosomes including 1B, 2D, 3A, 3B, 3D, 4A, 4D, 5A, 5B, 5D, 6D, and 7A. The chromosomes with more QTL loci were 5D, 5B, and 5A. Three hotspots of additive QTLs were detected on chromosomes 5A, 5B, and 5D. Seven genes related to seed vigor traits were screened from the hot spots on chromosome 5B, which were TRAESCS5B01G564900, TRAESCS5B01G564200, TRAESCS5B01G562600, TraesCS5B02G562700, TRAESCS5B01G561300, TRAESCS5B01G561400, and TRAESCS5B01G562100. These genes may be involved in the regulation of several processes including carbohydrate and lipid metabolism, transcription, and cell division, during the germination of aging seeds. These newly detected QTLs, SNP markers, and related candidate genes provide valuable information for molecular marker-assisted selective breeding of high-vigor wheat seeds.

Data availability
The experimental materials and relevant data involved in this paper can be used publicly.